EP4432137A1 - Method for operating a cable processing device, cable processing device, evaluation and/or control device for a cable processing device and machine-readable program code - Google Patents

Abstract

The invention relates to a cable processing device, an evaluation and/or control device for a cable processing device, machine-readable program code and a method for operating a cable processing device (100), comprising the method steps: processing (S1) at least one cable section (K, K1, K2, K3, Kn), in particular a cable end, by means of a cable processing unit (200) from an initial configuration to a final configuration (E, E1, E2, E3, En), capturing (S2) at least one image of the at least one final configuration (E1, E2, E3, En) by means of an image recording device (300), applying (S4, S5) a trained neural network to the at least one captured image, wherein the trained neural network is designed to, in a first step, select at least one region of interest (401, 402, 403, 404) of the at least one final configuration (E, E1, E2, E3, En) from the at least one image. to determine a captured image (S4) and, in a second step, to carry out a classification of the at least one determined region of interest (401, 402, 403, 404) with regard to the presence of at least one learned error image of the at least one final configuration (E, E1, E2, E3, En) and to determine a result (A1, A2, A3, A4) associated with the classification (S5), generating and outputting (S6) a control signal depending on the at least one determined region of interest (401, 402, 403, 404) and/or depending on the determined result (A1, A2, A3, A4) of the classification of the at least one final configuration (E, E1, E2, E3, En), whereby the cable processing process is further improved.

Description

Technical area

The invention relates to a method for operating a cable processing device, a cable processing device, a control device for a cable processing device and a machine-readable program code. The present invention is mainly described in connection with data cables. It is understood that the present invention can be used for any type of cable.

State of the art

When manufacturing data cables, the cables are usually processed in an automated processing system and, for example, assembled, i.e. cut to the appropriate length and provided with appropriate electrical contacts and/or plugs. A single automated processing system comprises several stations, which are also referred to as a cable processing unit in this application. In each cable processing unit, a defined processing step is carried out for a cable.

In order to enable error-free processing of such cables in large quantities, the cables in the respective processing plant must be checked to ensure that the individual processing steps have been implemented correctly. In particular, within a specific processing plant, it is necessary to check the cables fed to or leaving individual cable processing units before they reach this cable processing unit to see whether the cables meet the specified requirements in order to be able to be processed further. The cables usually have an initial configuration. This is the state in which a cable or a cable section to be processed is fed to a cable processing unit. The cable or cable section is then subjected to a processing step and transferred from the initial configuration to a final configuration. This is the state in which a cable or a cable section leaves the cable processing unit that processes the cable or cable section.

For this purpose, it is known to detect and evaluate cables or cable sections using image recognition.

Image processing programs that focus on classic rule-based object recognition are often inflexible and complex and sometimes very difficult to implement reliably.

Therefore, AI-based models are increasingly being used to monitor production. For example, the disclosure document EP 3 855 359 A1 a cable processing station is known which has an imaging sensor device by means of which a cable end can be detected. A first and a second cable-specific image parameter are recognized by means of an image processing system. On the basis of the recognized first and second cable-specific image parameters, a control-specific parameter is created by means of the image processing system, which is transmitted to a control device for controlling a tool. Such a method is intended to enable rapid tracking of the cable processing process for recognized cable types.

It has been shown that such processes also require further improvement, since the reliability and accuracy are often not sufficiently high for the problems faced in manufacturing.

Description of the invention

It is an object of the invention to provide a solution that creates a robust possibility to reliably and quickly determine a final configuration of a cable section and thus further improve the cable processing process.

The object is achieved by the subject matter of the independent claims. Advantageous developments of the invention are specified in the dependent claims, the description and the accompanying figures. In particular, the independent claims of a claim category can also be developed analogously to the dependent claims of another claim category. Further embodiments and developments emerge from the subclaims and from the description with reference to the figures.

The present invention relates in particular to a method for operating a cable processing device, comprising processing at least one cable section, in particular a cable end, by means of a cable processing unit from an initial configuration to a final configuration. In this case, at least one image of the at least one final configuration is captured by means of an image recording device. A trained neural network is applied to the at least one captured image, wherein the trained neural network is designed to determine at least one region of interest of the at least one final configuration from the at least one captured image in a first step and to second step, to determine a classification of the at least one determined region of interest with regard to the presence of at least one learned error pattern of the at least one final configuration and to determine a result associated with the classification. A control signal is then generated, depending on the at least one determined region of interest and/or depending on the determined result of the classification of the at least one final configuration, and output.

The present method provides a robust and fast solution by means of which learned final configurations of a cable section can be recognized and/or classified. In particular, the determination of the at least one region of interest and the classification can take place in real time, i.e. without delay in production.

It has been shown that the use of the same neural network for the determination of at least one region of interest and the subsequent classification of the determined region of interest and the associated result determination has proven to be particularly robust against variable environmental influences and variances, especially during image acquisition, and thus has a high level of reliability.

Furthermore, a neural network trained in this way can be constructed with fewer layers, which on the one hand means that the response times for any control of a process of the cable processing device are shorter and thus a control intervention can be carried out more quickly. Furthermore, the effort required to train the network is reduced. The trained neural network is advantageously designed as a deep neural network, also referred to as a deep neural network, in particular as a convolutional deep neural network, also referred to as a deep convolutional neural network.

The present invention therefore provides for the use of a specially trained artificial neural network and for the detection of error patterns to be carried out with the aid of the correspondingly trained neural network.

The neural network is trained using appropriate training data sets and developed as required. In the context of such training, the trainable neural network can be presented with appropriate training data, which have already been divided into positive training examples, e.g. final configurations having at least one error-free area of interest, and negative training examples, e.g. final configurations comprising at least one error-prone area of interest. The training data also includes also a corresponding result of the classification. Such a training data set can be created manually, for example, or from the results of a conventional image processing system in combination with the corresponding production information. The images are easy to obtain because cable production can be monitored routinely and the data is stored for some time. Furthermore, cable production is still monitored for errors and types of errors, mainly by personnel.

In the following, a possible embodiment for the creation of a training data set and the training of the neural network is described. It is understood that the described creation and training can also be carried out independently of a cable processing device. The present disclosure therefore explicitly discloses such creation and corresponding training as separate
objects.

A corresponding set of training images is generated for training the neural network. The images show at least one area of interest in a final configuration and an associated classification result, e.g. "OK" or "not OK". The training image data sets are qualified accordingly according to the area of interest shown and the classification result, whereby not only the criterion of error-prone vs. error-free can be provided, but the specific error pattern for the respective image can also be annotated.

The capture of corresponding images and their qualification can take place during normal production operations of the cable processing device, so that a large number of images with different sizes, viewing directions, rotation, contrast, lighting, occlusion, etc. can be generated and the area of interest, which is usually determined offline, and its classification or the classification result can be supplemented. This qualification of the images can be done manually or at least partially by a conventional image processing system, whereby manual rework is possible. The pre-qualified training images are then pre-processed accordingly for training.

For example, the size of the training images can be adjusted to a predetermined size. In particular, the number of pixels of the training images can be adjusted to the number of inputs of an input layer of the neural network. The preprocessing of the training images can also include normalizing the images. For example, the images can be available as RGB images as they are transmitted by an image recording device. This means that preparation and preprocessing of the images are preferably avoided so that the trained neural network can handle corresponding raw data, which increases the subsequent speed of image evaluation.

In order to improve the results of the neural network and make them more robust, the training images can be subjected to random image manipulations. Such image manipulations can include, for example, rotating, enlarging, reducing and/or distorting. It goes without saying that corresponding orders of magnitude can be specified for the respective image manipulation.

For example, a maximum enlargement or reduction in percent can be specified for enlarging or reducing, e.g. 110% or 90%. For rotation, a maximum or minimum angle of rotation can be specified, e.g. +/- 10°, 20° or 30°. Corresponding limit values can also be specified for distortion. It goes without saying that different algorithms can be used for image distortion, which can have different parameters.

The image manipulations serve to provide the training data with greater variability. The final configuration or the majority of final configurations will therefore be present in different places and in different sizes in the image. This prevents, for example, the neural network from learning to identify the given feature only in a small section of an image and falsely concluding that the feature is not present, even though it is just outside the section.

After preprocessing the training images, the neural network is trained with a portion of the training images obtained in this way. The remaining training images can be used to check the learning success by feeding them to the trained neural network and comparing its output with the known or expected output for the respective training image. This portion of the training images can therefore also be referred to as test data.

After training has been completed, the quality of the training can be checked, as mentioned above, by qualifying the test data using the trained neural network. If the results reach the desired quality, training can be ended. If the results have not yet reached the desired quality, or if the neural network needs to be further developed, training can be continued with the appropriate training data or carried out again with changed parameters.

The at least one final configuration is recorded using at least one image from an image recording device. The at least one image is preferably recorded digitally, for example using a CCD camera to generate images in a spectral range visible to the human eye. The image recording device is positioned and aligned in such a way that the at least one final configuration is typically arranged in the recording area of the image recording device. If necessary, a plurality of images or a plurality of image recording devices can also be provided which reliably record the at least one final configuration. It can be more advantageous to enlarge the recording area and to adapt the resolution of the image recording device accordingly.

It goes without saying that training can be carried out differently depending on the type of neural network used. Basically, with any type of training, the weights of the neural network are adjusted with each training run so that the error of the neural network's output is minimized compared to the known result from the training data set. This is usually achieved by so-called back-propagation.

A so-called number of epochs and a termination criterion can also be specified for training. The number of epochs indicates the number of training runs. A specified amount of training data, e.g. all training data or just a selection of training data, can be used for each training run. The termination criterion indicates how far the result of the trainable neural network can deviate from the ideal result in order for the training to be considered successfully completed and for the neural network to be considered sufficiently well trained.

In particular, the neural network can be designed as a deep convolutional neural network, abbreviated to dCNN. CNNs and dCNNs deliver very good results, especially for classifying objects in image data. It goes without saying that other suitable neural networks from the field of machine learning are also possible.

Such a dCNN can have an input layer as well as a plurality of hidden layers and an output layer. The hidden layers can have at least partially identical or repeating layers.

The input layer can have an input for each pixel of the captured images. It is understood that the images can be stored as an array or vector with the corresponding number of elements can be transmitted to the input layer. Furthermore, the size of the captured images can be the same for all images. For example, the images can be captured with 1024*1024 pixels, which represents a capture area of 25cm by 25cm for corresponding final configurations. It is understood that these details are only examples and other image parameters can be used.

A schematic layer structure of the neural network is explained below. The image data to be analyzed is fed to the input layer of the neural network using the first layer of the trained neural network.

Layer 1: Input Tensor

The hidden layers can carry out a kind of preparation of the input data for further processing by the neural network and then process it further in a large number of identical blocks. These layers can be, for example, layers for filling with zeros, so-called zero padding layers, layers for convolution, layers for normalization, and layers for activation, in particular by means of a so-called ReLU function, also called a rectified linear unit.

• Schicht 2: Transpose (Transponierung des input Tensors zur Weiterverarbeitung)
• Schicht 3: Sub (Subtraktionsoperation für transponierten Tensor)
• Schicht 4: Faltung (Faltungsoperation)
• Schicht 5: Aktivierung (Aktivierungsfunktion - ReLU)
• Schicht 6: MaxPooling (Selektion von Maximalwerten)

An example layer structure of such layers for processing the data can be as follows:

• Layer 2: Transpose (transposing the input tensor for further processing)
• Layer 3: Sub (subtraction operation for transposed tensor)
• Layer 4: Convolution (convolution operation)
• Layer 5: Activation (Activation Function - ReLU)
• Layer 6: MaxPooling (selection of maximum values)

• Schicht 7: Faltung (Faltungsoperation)
• Schicht 8: Aktivierung (Aktivierungsfunktion - ReLU)
• Schicht 9: Faltung (Faltungsoperation)
• Schicht 10: Aktivierung (Aktivierungsfunktion - ReLU)
• Schicht 11: Faltung (Faltungsoperation)
• Schicht 12: Addition (Additionsoperator)

Another example block for further analysis of the processed data can follow:

• Layer 7: Convolution (convolution operation)
• Layer 8: Activation (Activation Function - ReLU)
• Layer 9: Convolution (convolution operation)
• Layer 10: Activation (Activation Function - ReLU)
• Layer 11: Convolution (convolution operation)
• Layer 12: Addition (addition operator)

Data from layer 6 can also be processed in parallel with just one convolution operation, which is then added to the result of layer 11 in layer 12. It goes without saying that different layer arrangements are possible and the above structure is shown only as an example.

These layers can also be followed by blocks, each of which can have an identical structure to the previously shown block.

Using the output layer of the neural network, corresponding information on the area of interest and the classification can then be output, e.g. number of areas of interest determined, type of areas of interest determined (e.g. crimp sleeve, grommet, internal contact, etc.), error status, such as faulty or error-free, a corresponding probability for the existence of the determined error status and identification of a specific error pattern.

As a result, a neural network with, for example, 274 layers can be provided, which is trained by means of corresponding training data to carry out the image evaluation in the manner according to the invention.

The end configurations can be designed in different ways. Typically, the end configuration of a cable section is characterized by the preceding processing step which takes place between the initial configuration and the end configuration. In particular, the cable section can be a cable end.

However, it is possible that the entire cable is also considered to be the final configuration, in particular if different parts of the cable are processed in different cable processing units and different cable sections of the same cable are captured with different image capture devices. The final configuration of the cable is always changed when a cable section of this cable is further processed.

In addition to the cable section, the final configuration can comprise at least one further element, for example from the group of crimp sleeve, contact pin, plug, inner conductor, outer conductor, shielding braid, connecting tube, cable, cable sheath, insulator, grommet, laser marking of the cable as an alphanumeric character string, coupling, holder for grommets, cable holder, separator, housing, dirt flap, ferrule for optical conductors, label, silicone insulating tube, etc., which are processed in the processing step of the processing unit together with the cable section. The final configuration can also comprise only elements of the cable, e.g. a stripped end with an adjacent non-stripped part as part of a partial stripping processing step. The final configuration can also only include components arranged on the cable, but not the cable itself, e.g. the combination of housing and housing dirt flap.

Since at least one cable section, in particular one cable end, is processed in the cable processing unit and a final configuration is always associated with the processed cable section, a corresponding plurality of final configurations can also be present when a plurality of cable sections are processed simultaneously by the cable processing unit. In this sense, the presence of at least one cable section means that at least one final configuration is present.

As a rule, the majority of the final configurations of the cable sections that are processed simultaneously by the same cable processing unit are similar; e.g. all cable sections, e.g. cable ends, are each fitted with a crimp sleeve and crimped. This means that the final configuration, here the crimp sleeve crimped onto the cable section, differs for the majority of cable sections generally only in terms of whether or not there is an error pattern for the respective final configuration of the respective cable section, e.g. whether it was crimped incorrectly or not.

The region of interest in the final configuration of at least one cable section is determined by means of the trained neural network as the region of interest which was learned as the region of interest according to the training data, provided that this region is present in the captured image. The region of interest in the final configuration can in particular be selected from the group: crimp sleeve, contact pin, plug, inner conductor, outer conductor, shielding braid, connecting pipe, cable, cable sheath, insulator, grommet, laser marking of the cable as an alphanumeric character string, coupling, holder for grommets, cable holder, separator, housing, dirt flap, ferrule for optical conductors, label, silicone insulating hose, etc. If a plurality of final configurations are present because a plurality of cable sections are processed by means of the cable processing unit, a plurality of regions of interest are determined accordingly, whereby only one region of interest or a plurality of regions of interest can be determined for each final configuration.

It is often sufficient to identify only one area of interest per final configuration and to classify it using the trained neural network. In this case, an isolated classification of the identified area of interest is carried out. However, several regions of interest for a particular final configuration. This is particularly of interest for different components, which may include the cable section, and comprehensive final configurations.

As a result of the classification, it can be determined in a first step, for example, whether a final configuration is faulty or error-free. For example, the result can be "OK" or "OK" for a final configuration that was determined to be error-free or "not OK" or "not OK" for a final configuration that was determined to be faulty. In particular, the result can be communicated using color codes, e.g. green marking of an area of interest as error-free and red marking of an area of interest as faulty. In a second step, e.g. at the request of a user or a control device, the specific error pattern can be identified from a large number of possible error patterns for faulty final configurations. The complete information of the output vector of the neural network can also always be made available as standard, especially for a control device.

In a further embodiment of the method, an image is captured by means of the image capture device which shows two to 100 final configurations, wherein the trained neural network is designed to determine at least one region of interest for the two to 100 final configurations, in particular for each of the two to 100 final configurations, and to determine at least one result associated with the classification of the regions of interest, and the trained neural network is applied to the captured image, wherein a control signal for the two to 100 captured final configurations, in particular for each of the two to 100 captured final configurations, is generated and output on the basis of the determined at least one region of interest and/or the result associated with the classification for the respective final configuration.

Such a design makes it possible to significantly increase the throughput of a cable processing unit through parallel processing, without running the risk that efficient and reliable error pattern control by a human supervisor is no longer possible. Rather, the use of an appropriately designed neural network enables reliable error pattern control in real time for a large number of cable sections or end configurations processed in parallel, thus significantly improving the production process. By generating and outputting a corresponding control signal, the corresponding cable or cable section can be further processed depending on the classification result, for example it can be specifically excluded from further parallel processing.

Preferably, the captured image shows at least 10 to 100, in particular 20 to 100, in particular 30 to 100, in particular 40 to 100 final configurations. For each final configuration shown, at least one region of interest is determined by means of the neural network trained for this purpose.

In a further development of the aforementioned embodiment, the trained neural network is designed to determine at least one region of interest with an associated probability value for a maximum number of final configurations, wherein the neural network is further designed to determine a probability value below a predetermined threshold value for a difference between the number of final configurations recorded and the maximum number, wherein the neural network is further designed so that regions of interest with a probability value below the predetermined threshold value are not classified. The neural network can therefore be used flexibly for a variable number of cables processed in parallel. Furthermore, no classification is carried out for regions of interest whose probability value does not exceed a threshold value because they are not present in the recorded image, which increases the speed of the neural network and saves computing resources.

In a further embodiment of the method, the trained neural network is designed to determine at least one first region of interest for at least one final configuration and at least one second region of interest that is spatially different from the first region of interest for the same final configuration from the at least one captured image, and the trained neural network is further designed to additionally take into account a relative parameter of the at least first region of interest and the second region of interest for the classification and to determine a result associated with the classification, and wherein the trained neural network is applied to the at least one image, wherein a control signal for the at least one final configuration is generated and output based on the result of the classification taking the relative position into account.

Thus, a plurality of different regions of interest, in particular two, three or four regions of interest, can be determined for a final configuration. Furthermore, a plurality of different regions of interest can be determined for a plurality of final configurations, in particular 2 to 100. In addition, it allows a relational Relationship between different cable sections of the same cable, here taking into account the final configuration, in addition to the final configurations of the cable sections, across different cable processing units.

In this embodiment, the classification is no longer carried out in isolation for the determined region of interest, but a relative parameter of the at least two determined regions of interest of the same final configuration is also taken into account for the classification of the final configuration. The relative parameter can be, for example, the relative position. Relative position is to be understood broadly in this context, and can include, for example, a relative orientation, a relative displacement or relative rotation or even a distance between the at least first and at least second regions of interest.

By including the relative position of the at least first and at least second region of interest in the classification, information can be obtained as to whether different components covered by a final configuration are arranged correctly relative to one another. For example, a crimp sleeve can be identified as the first region of interest and a stripped cable end as the second region of interest, with the distance between the first region of interest and the second region then being assessed. From this, it can be determined whether the crimp sleeve is positioned correctly on the stripped cable end, in particular whether the crimp sleeve covers the stripped cable end too far or is arranged too far away from the stripped region. Furthermore, the longitudinal direction of the crimp sleeve and the longitudinal direction of the stripped cable end can also be taken into account in order to determine whether the crimp sleeve is at an angle relative to the cable end.

Furthermore, a functional affiliation of the first and second areas of interest to a specific component class can be taken into account as a relative parameter. For example, it can be determined whether a grommet, the first area of interest, is arranged at one end of an almost completely processed cable and whether a crimp sleeve is provided at the other end, the second area of interest, which is still being processed.

In this case, the neural network is trained in such a way that a determined region of interest is associated with a specific component, e.g. that a first region of interest is a grommet and a second region of interest is a crimp sleeve, whereby a comparison of the number of grommets and crimp sleeves for the same cable is then taken into account as a relative parameter for the classification. Furthermore, the neural network can be designed to determine whether for each determined region of interest in the form of a Crimp sleeve has exactly one area of interest in the form of a grommet along the cable. It is understood that this is only an example and the classification, which takes a relative parameter into account, can also be provided for other components.

This embodiment allows complex final configurations assembled using different components to be classified in real time and thus the manufacturing process to be handled reliably. In particular, this form of classification can be applied to an image comprising 2 to 100 final configurations, so that complex processing processes can be monitored for errors at high throughput.

In a further embodiment, the image of the at least one final configuration is captured using a digital image recording device such that at least one region of interest is represented by at least 20 pixels, preferably more than 50 pixels. It has been shown that the trained neural network delivers particularly reliable results if a region of interest is represented by at least 20 pixels, preferably more than 50 pixels in the captured image. This can be achieved, for example, if the image recording device has a resolution of 1 megapixel or more and the recording area in which the plurality of final configurations is arranged is 25cm by 25cm.

In a further embodiment, the trained neural network is designed to determine the complete image capture of a final configuration, wherein the trained neural network is applied to the captured image, wherein in the case of the determined incomplete capture of at least one final configuration, a control signal is generated and output which causes an image capture error to be reported. This is generally not an error in the manufacturing process, but an error in the image capture process. This does not mean that processed cable sections that were not captured correctly are faulty. It simply means that an area of interest and/or a classification cannot be carried out for this final configuration using the captured image due to the incomplete capture of a certain final configuration. In the event of an error, a check must therefore be made as to whether the image capture device needs to be readjusted or whether a final configuration was not fully captured for other reasons, e.g. defective cable routing. Preferably, the trained neural network for determining the completeness of a final configuration is the same neural network which is also designed to perform the at least one region of interest and the classification of the at least one region of interest.

In a further embodiment, the control signal causes a graphic display of the at least one determined region of interest in an image showing the at least one final configuration, in particular the image to which the trained neural network was applied, in particular as a colored border of the at least one determined region of interest, on an image output device. By outputting a graphic display on an image output device, it becomes clear to the staff whether and which region or regions in the captured image were determined as the region or regions of interest. This allows a plausibility check to be carried out. Preferably, a colored frame or a colored border can be used to make a determined region of interest or determined regions of interest quickly identifiable for the staff on the image showing the final configuration.

Advantageously, the frame can identify the type of area of interest determined, e.g. a specific cable section or a specific component, by means of the frame color.

According to a further embodiment, the control signal is used to cause a graphic display of the result of the classification for the at least one final configuration in an image showing the at least one final configuration, in particular the image on which the classification is based, on an image output device. The graphic display of the result of the classification can in particular be reproduced as a color index and/or as a text block, e.g. "OK" or "OK" or "not OK" or "not OK".

In particular, it can be advantageous if the frame indicates by means of the color for which at least one determined region of interest there is a classification result that corresponds to an error image and for which at least one determined region of interest there is a classification result that corresponds to an error-free state.

This graphical output also allows the staff to quickly and easily visually determine which final configuration or which areas of interest were classified as faulty or fault-free.

In a further embodiment, the graphic display shows an error pattern associated with the result of the classification in the form of an error pattern name. This not only enables the staff to determine whether a final configuration is faulty or error-free in the sense of "OK" or "not OK", but the staff is also informed about the type of error pattern.

In a further embodiment of the invention, the operation of the cable processing unit is influenced by means of the control signal in such a way that the final configuration of a cable section to be processed subsequently is brought closer to a desired, in particular error-free, final configuration by means of the same cable processing unit. In this case, the result of the classification is used for a control intervention of the cable processing unit in order to avoid, as far as possible, a corresponding error pattern for final configurations subsequently produced by means of the cable processing unit.

In this context, the determined region of interest and the result of the classification can be used as input data for a trained neural network for controlling the cable processing unit. This trained neural control network is preferably designed to generate and output a control signal for influencing at least one control variable of the cable processing unit based on at least one determined region of interest and an associated classification result, which reduces the probability of the presence of an incorrect final configuration after processing for a next cable section to be processed.

In a further embodiment, the control signal is used to influence a cable sequence processing of a cable sequence processing unit downstream of the cable processing unit of at least one classified final configuration depending on the result of the classification of this final configuration. This allows corrective processing of faulty final configurations by subsequent processes, as far as possible, in order to avoid production waste.

In this context, the determined region of interest and the result of the classification can be used as input data for a trained neural network for controlling a subsequent process, i.e. a processing process that follows the processing step in the context of which the final configuration was determined to be faulty. This trained neural control network is preferably designed to generate and output a control signal for influencing at least one control variable of a cable sequence processing unit based on at least one determined region of interest and an associated classification result. This allows an incorrect final configuration to be processed individually in a corrective manner, so that production waste is avoided.

In a further embodiment, the control signal is used to cause a cable section with a faulty final configuration to be rejected from a production process that is still to be completed, whereby rejection takes place when the result of the Classification for this final configuration corresponds to a given error pattern with a probability above a given probability threshold and cannot be corrected by means of a subsequent cable processing unit. This alternative is advantageous if an error pattern of a final configuration cannot be corrected by subsequent processes. This can be the case for certain given error patterns that have been identified with a sufficiently high degree of certainty. These error patterns that cannot be corrected and are present with a high degree of certainty can depend on the respective production structure and can also depend on the arrangement of the cable processing units in the direction of material flow. If the error pattern of the final configuration cannot be corrected, it is advantageous to stop processing the faulty final configuration and to remove or separate it from the production process as soon as possible.

The invention also relates to a cable processing device comprising:
a cable processing unit which is designed to receive at least one cable section, in particular a cable end, and to process it in such a way that the at least one cable section is transferred from an initial configuration to a final configuration, an image recording device, arranged and designed to record at least one image of the at least one final configuration, an evaluation device and a control device for controlling and/or regulating the cable processing device, wherein the evaluation unit is operatively connected to the image recording device and the control device, and wherein machine-readable program code can be loaded into the evaluation device and/or the control device, which, when executed, causes the method according to one of the preceding claims to be carried out. By means of such a cable processing device, in particular an automated quality check of the cable production process can be carried out.

The cable processing device thus comprises in particular a trained neural network for evaluating the at least one recorded image of the at least one final configuration, wherein the neural network is trained such that in a first step at least one region of interest of the at least one final configuration is determined from the image and in a second step a classification of the at least one determined region of interest takes place with regard to the presence of an error image of the final configuration, wherein the evaluation device is designed to output a control signal depending on the determined region of interest and/or depending on a determined result of the classification of the final configuration.

The invention also relates to a machine-readable program code for an evaluation device and/or control device, which comprises control commands which, when executed by the evaluation device and/or the control device, cause the method according to one of the method claims to be carried out.

The invention also relates to a control and/or evaluation device with machine-readable program code, which comprises control commands which, when executed by means of the evaluation device and/or control device, cause the method according to one of the method claims to be carried out.

The evaluation device advantageously accesses a trained neural network, in particular in the form of a machine-readable program code, which is designed to determine at least one region of interest of the at least one final configuration from the at least one captured image in a first step and, in a second step, to carry out a classification of the at least one determined region of interest with regard to the presence of a learned error image of the at least one final configuration and to determine a result associated with the classification, and to generate and output a control signal depending on the at least one determined region of interest and/or depending on the determined result of the classification of the at least one final configuration. The evaluation device can access, in particular the program code implementing the trained neural network, by accessing a local memory; alternatively, access can also be by accessing a cloud on which the corresponding neural network is implemented.

Advantageously, the trained neural network, which the evaluation device accesses, is further designed to determine at least one region of interest for two to 100 final configurations, in particular for each of the two to 100 final configurations, and to determine a result associated with the classification of the regions of interest, wherein a control signal for the two to 100 recorded final configurations, in particular for each of the two to 100 recorded final configurations, can be generated and output on the basis of the determined at least one region of interest and/or the result associated with the classification for the respective final configuration.

Advantageously, the trained neural network, which the evaluation device accesses, is further designed to identify at least one first region of interest for at least one final configuration and at least one second region of interest for the same final configuration, which region is to determine a spatially different region of interest from the at least one captured image and the neural network is further designed to additionally take into account a relative parameter, in particular a relative position, of the at least first region of interest and the second region of interest for the classification and to determine a result associated with the classification, wherein a control signal for the at least one final configuration can be generated and output based on the result of the classification taking the relative position into account.

Short character description

Figur 1

eine schematische Darstellung einer Ausführungsform der Kabelverarbeitungsvorrichtung,

Figur 2

eine andere schematische Darstellung der Ausführungsform aus Figur 1,

Figur 3

eine schematische Darstellung der Ausführungsform der Kabelverarbeitungsvorrichtung geeignet für eine übergreifende Beeinflussung der Kabelfertigung,

Figur 4

eine erste beispielhafte Endkonfiguration für welche eine Mehrzahl an interessierenden Bereichen und deren Klassifikation ermittelt wurde,

Figur 5

eine zweite beispielhafte Endkonfiguration für welche eine Mehrzahl an interessierenden Bereichen und deren Klassifikation ermittelt wurde,

Figur 6

eine dritte beispielhafte Endkonfiguration für welche eine Mehrzahl an interessierenden Bereichen und deren Klassifikation ermittelt wurde,

Figur 7

eine vierte beispielhafte Endkonfiguration für welche eine Mehrzahl an interessierenden Bereichen und deren Klassifikation ermittelt wurde,

Figur 8

ein Flussdiagramm zur schematischen Darstellung eines beispielhaften Ablaufs des Verfahrens.

Advantageous embodiments of the invention are explained below with reference to the accompanying figures. They show:

Figure 1

a schematic representation of an embodiment of the cable processing device,

Figure 2

another schematic representation of the embodiment of Figure 1 ,

Figure 3

a schematic representation of the embodiment of the cable processing device suitable for a comprehensive influence on cable production,

Figure 4

a first exemplary final configuration for which a plurality of regions of interest and their classification were determined,

Figure 5

a second exemplary final configuration for which a plurality of regions of interest and their classification were determined,

Figure 6

a third exemplary final configuration for which a plurality of regions of interest and their classification were determined,

Figure 7

a fourth exemplary final configuration for which a plurality of regions of interest and their classification were determined,

Figure 8

a flow chart schematically showing an exemplary procedure of the method.

The figures are merely schematic representations and serve only to explain the invention. Identical or equivalent elements are provided with the same reference numerals throughout.

Detailed description

Figure 1 shows a schematic view of a cable processing device 100. The cable processing device 100 comprises at least one cable processing unit 200, by means of which a cable or a cable section K is processed from an initial configuration into a final configuration E. The initial configuration refers to the state of the cable section K to be processed before processing. The final configuration refers to the state of the processed cable section K.

It is understood that a plurality of cable processing units 200 can be included in the cable processing device 100, each of which carries out different processing steps on the cable, possibly also on different cable sections of the same cable.

The final configuration E can depend on the processing status of the cable or the cable section K as well as on the type of cable to be manufactured. In Figure 1 the final configuration comprises the cable section K of the cable designed as an end region and a schematically shown crimp sleeve C arranged thereon. The final configuration can be imaged in the flow direction after the processing step, possibly also after exiting the cable processing unit 200, for example on a cable receiving device 210.

This final configuration E is recorded by means of an image recording device 300. The recording area of the image recording device 300 for recording final configurations E is preferably 25 cm by 25 cm. If a final configuration E is arranged within the recording area, the image recording device 300 can record the final configurations E arranged therein in the recording area.

The image recording device 300 is preferably designed as a digital camera and has a resolution of 1024x 1024 pixels. With this configuration of recording area and resolution, it is ensured that the image has a sufficiently good resolution to carry out a reliable image evaluation. A lower resolution with the same recording area or a larger recording area with a smaller resolution can have a negative impact on the result of the image evaluation. It should be ensured that an area of interest has at least 20 pixels, preferably 50 or more pixels. However, a larger recording area and a higher resolution can be selected, although this may have disadvantages in terms of the speed of subsequent image processing.

The cable processing device 100 further comprises an evaluation device 500. An image evaluation is carried out by means of the evaluation device 500. The evaluation device 500 comprises a trained deep convolutional neural network, which is designed to determine at least one region of interest of the at least one final configuration E from the at least one captured image in a first step and to determine a classification of the at least one determined region of interest with regard to the presence of a learned error image of the at least one final configuration E and a result associated with the classification in a second step.

For this purpose, a machine-readable code 600 is loaded into a non-volatile memory of the evaluation device 500, which includes the trained neural network. By applying the trained neural network to the image, the evaluation device 500 can determine at least one region of interest of the final configuration E and classify this at least one region of interest.

Preferably, at least one region of interest is determined for each final configuration E captured by means of the image. It is also possible to determine several regions of interest per final configuration E.

Based on the result of the classification for the determined areas of interest, the evaluation device 500 sends a control signal to a control device 700 included in the cable processing device 100. This control device 700 is designed to control or regulate the cable processing device 100. The evaluation device 500 can be provided separately from the control device 700. The evaluation device 500 can also be integrated in the control device 700.

Furthermore, the control device 700 is also operatively connected to an image output device 800. The control device 700 is designed to cause the information provided by the evaluation device 500 to be displayed on the image output device 800. The determined regions of interest and the respective classification results are preferably displayed on the image comprising the analyzed final configuration E. In the input layer, the appropriately trained deep convolutional neural network is provided with raw data from the image recording device 300 that has not been preprocessed or otherwise prepared, for example in the form 1024x1024x3. A tensor is therefore provided that has 1024x1024 pixels, each of which can have 3 RGB colors. The aim is that the captured images can be evaluated directly using the evaluation device 500 without any further intermediate processing or preparation.

The output data can include, for example, the maximum number of regions of interest in the image, associated probability values for each of the regions of interest, the localization of the regions of interest by means of frames in the image, and the classification result, which distinguishes between those with errors and those without errors, and possibly other information, which can include, for example, the type of error pattern and a further probability value for the presence of a specific error pattern. An output vector is thus output for the intended number of regions of interest that can be determined using the neural network, with the respective information on the respective regions of interest.

The deep convolutional neural network used is designed to initially determine at least one region of interest per final configuration E of a cable section K, preferably several regions of interest per final configuration E of a cable section K, for a large number of cable sections using the same deep convolutional neural network and then to classify the determined regions of interest. This step-by-step procedure is carried out using the same network that was trained for this procedure.

Figure 2 illustrates the embodiment according to Figure 1 with regard to the parallel cable processing capacity of the cable processing unit 200. The cable processing unit 200, as well as the cable processing device 100, is designed to process a large number of cables, here n cables, in parallel. Preferably, 30 to 100 cables are processed simultaneously by the cable processing unit 200 and each is converted from an initial configuration to a final configuration.

This means that the challenge is to provide real-time control for a large number of cables, eg 30 to 100 cables, to determine whether each individual cable has been processed without errors. The n end configurations are designated E1, E2, E3 to En. The corresponding cable sections are designated K1, K2, K3 to Kn.

It has been shown that the two-stage procedure, according to which at least one region of interest is determined for at least one final configuration and the determined at least one region of interest is then classified, is particularly successful even for a large number of cable sections processed in parallel. This requires a neural network trained with appropriate training data.

The neural network is preferably designed in such a way that it is configured to determine up to 100 regions of interest from 100 final configurations E1, E2, E3, ..., En processed in parallel using the same cable processing unit 200. It is also configured to provide the 100 regions of interest with probability values. If fewer than 100 final configurations are processed simultaneously, it determines a probability value for the difference between the number of final configurations recorded and 100, which is below a predetermined threshold value. The regions of interest with a probability below this threshold value are no longer taken into account and are not classified by the neural network.

This provides a neural network that is suitable for handling a flexible cable throughput through the cable processing unit up to a maximum number of cables or maximum number of final configurations E1, E2, E3, ..., En in real time and for checking the production process. The maximum number of areas of interest that can be determined using the neural network can be selected such that it corresponds to the maximum parallel processing capacity of the cable processing unit 200. For example, this can be selected to be a factor of 2 to 5 higher than the number of cables that can be processed in parallel at most, so that a plurality of areas of interest can also be determined for each final configuration E1, E2, E3, ..., En.

For the multitude of final configurations E1, E2, E3, ..., En, several regions of interest can be determined, which allows a relational classification that takes into account relative parameters of two or more regions of interest to each other.

Figure 3 shows a cable processing device 100. This comprises the already Figure 1 and 2 shown cable processing unit 200, as well as a cable follow-up processing unit 201 downstream of the cable processing unit in the flow direction, and a cable pre-processing unit 202 upstream of the cable processing unit 200 in the flow direction. Furthermore, the cable processing device 100 comprises a higher-level control device 710, by means of which the cable pre-processing unit 202 and the cable subsequent processing unit 201 can be influenced at least indirectly. The flow direction or material flow direction is designated by D.

While the control device 700 is designed to control or regulate the operation of the cable processing unit 200, the higher-level control device 710 is at least indirectly suitable for influencing the operation of at least one cable pre-processing unit 202 upstream of the cable processing unit 200 in the flow direction D. Furthermore, the higher-level control device 710 is also at least indirectly suitable for influencing the operation of at least one cable subsequent processing unit 201 downstream of the cable processing unit 200 in the flow direction D.

The cable pre-processing unit 202 and the cable subsequent processing unit 201 can be units directly preceding or following the cable processing unit 200. However, they can also be further away from the cable processing unit 200.

The at least one cable pre-processing unit 202 and the at least one cable sequence processing unit 201 can each have their own control device for controlling and/or regulating their operation. Such a control device is shown in Figure 3 not shown. It is understood that an influence on a cable pre-processing device 202 and a cable subsequent processing device 201 does not necessarily have to be provided. Rather, only an influence on the cable pre-processing device 202 or the cable subsequent processing device 201 can be provided.

On the basis of the image of the final configurations E captured by the image recording device 300, the evaluation device 500 determines at least one region of interest of the final configurations E and classifies the at least one region of interest. If the classification of a region of interest leads to the result that an error pattern exists for a specific final configuration E and what type of error pattern it is, the cable processing device 100 can be controlled and/or regulated, for example in the form of the control or regulation of the cable sequence processing device 201, such that for the faulty final configuration E, corrective processing of this final configuration E takes place in a subsequent processing step by means of the cable sequence processing unit 201.

If, for example, a crimp sleeve is not pressed firmly enough onto a stripped cable end in the cable processing unit 200, the position of the crimp sleeve is correct, but this is also recognized as incorrectly pressed using the classification. This error pattern is repairable. The error pattern for this final configuration can be corrected, for example, when the housing is applied by applying the housing that holds the crimp sleeve with a correspondingly increased contact force to the final configuration classified as repairably faulty. This can mitigate or compensate for the incorrect pressing of the crimp sleeve so that the cable produced is ready for delivery and meets the quality standards.

This approach can generally be applied to all classified areas of interest that are known to have a repairable defect pattern that can be adequately corrected by downstream processing steps. This avoids waste, achieves greater manufacturing efficiency and reduces the cost per cable.

Furthermore, the control device 700 is also designed to control and/or regulate the cable processing unit 200 based on the at least one determined region of interest and the associated classification result for a specific final configuration E in such a way that the determined error pattern is reduced or completely avoided for subsequently produced final configurations.

In the example given with the crimp sleeve, for example, the contact pressure of the crimping press can be increased, in particular only for the position of the cable in the cable processing unit 200 for which the fault pattern was detected. The determined area of interest and the associated classification can therefore be used to individually adapt the control variables of the cable processing unit 200 in such a way that a stable and optimized processing process is achieved for subsequent cable sections.

Furthermore, the cable processing device 100 can be made of Figure 3 be designed to influence the operation of a cable pre-processing unit 202 based on the at least one determined region of interest and the associated classification result in such a way that a specific error pattern in the final configuration E of the cable section K, detected after processing with the cable processing unit 200, is reduced. This can also be done by means of the higher-level controller 710.

For example, the evaluation may show that the cable has not been stripped correctly and, for this reason, the crimp sleeve applied in the cable processing unit 200 is not positioned correctly. In this case, the error is caused by the cable pre-processing unit 202, which carries out the step of stripping the cable section. Consequently, this error can be remedied if the operation of the cable pre-processing device 202 is influenced in a way that changes the stripping process back towards a desired result.

It is understood that these are merely exemplary explanations. Depending on the determined area of interest and its classification, this can be extended to any final configuration of the cable section. This applies accordingly to cable pre-processing unit 202 and cable subsequent processing unit 201.

This approach makes it possible to avoid the need for an image capture device to record final configurations and evaluate the image after each cable processing unit. Instead, a few image evaluations at a few cable processing units are sufficient, and the results can be used to positively influence a large number of production processes.

Figure 4 shows a first example final configuration E of a cable section K. The final configuration E comprises, from left to right, a sheath M of the cable, followed by a shield S of the cable, the shield S is followed by a crimp sleeve C. The crimp sleeve C is again followed by a sheath M of the cable.

Such a final configuration E can be captured by means of an image from the image processing device and is fed to the evaluation device. Figure 4 now shows an exemplary result of an application of the neural network included in the evaluation device to the captured image of the final configuration E. Accordingly, this can be done if the image shows a plurality of final configurations E.

The evaluation device is used to determine four regions of interest for the current final configuration E. A first region of interest 401 is determined, which is assigned to the jacket on the left in the figure. This has a probability above a threshold value, so that R1 is marked using a box or a frame.

The neural network provides a position of the frame as an xy coordinate in the image with an associated length in x and y directions. This applies to all frames in Figure 4 and also the following characters accordingly.

Furthermore, a second region of interest 402 is determined by means of the neural network, which has a probability value that is above a threshold value. This second region of interest is assigned to the shielding S and is also marked with a frame R2 on the image.

Furthermore, a third region of interest 403 is determined by means of the neural network, which has a probability value that is above a threshold value. This third region of interest is assigned to the crimp sleeve and is also marked with a frame R3 on the image.

Furthermore, a fourth region of interest 404 is determined by means of the neural network, which has a probability value that is above a threshold value. This fourth region of interest is the region of interest in the Figure 4 right-hand coat M and is also marked with a frame R4 in the picture.

The frames R1, R2, R3, R4 are preferably assigned information A1, A1, A2, A3, which contain further information on the determined region of interest 401, 402, 403 and 404. For example, the information A1 to A4 can each include the probability with which a region of interest was determined. Furthermore, these pieces of information A1, A1, A2, A3 preferably include the result of the classification, as set out above.

In particular, it can be advantageous for staff to be able to identify the result of the classification more quickly, e.g. OK vs. not OK, as a color coding of the frames R1 to R4. For example, a traffic light system can be used, with a green frame representing error-free and a high probability of error-free, red representing error-prone and a high probability of error, and yellow representing classification results with insufficiently high probability, meaning that they require checking.

Furthermore, the information A1 to A4 can, if necessary only upon user request, e.g. by hovering over the information using a control device, e.g. a mouse, to specifically identify the error pattern, so that the staff can quickly determine and see which error pattern is present for a final configuration E.

Each of these determined areas can be classified in isolation, e.g. whether the shielding braid has been processed in accordance with the process or not; whether the crimp sleeve C has been processed in accordance with the process or not, in particular whether this is the case in the pressed or unpressed state or whether the pressing was carried out without errors.

In the present final configuration E, a further relational component is taken into account in the classification, namely the crimp sleeve C, in Figure 4 the third area of interest 403, correctly between the shielding braid S, in Figure 4 the second region of interest 402, and the jacket M, in Figure 4 the fourth region of interest 404. This means that the relative position of the third region of interest 403 to the second and fourth regions of interest 402 and 404 is error-free.

It can thus be determined by means of the evaluation device for this final configuration E whether the crimp sleeve C is arranged in the correct position, namely between the shield braid and the jacket and whether the crimp sleeve has been correctly pressed or has not been correctly pressed or has not been pressed.

Such an exemplary evaluation can be carried out for a large number of final configurations E that are processed in parallel and captured in images.

Figure 5 shows a second exemplary final configuration E of a cable section. This final configuration E shows a cable K, determined as the first area of interest 401, and a connecting pipe V, determined as the second area of interest 402. The determined areas of interest 401 and 402 are again identified by means of a frame R1 or R2. The corresponding information A1 or A2, such as for Figure 4 explained, are arranged in the corresponding frame R1 and R2.

Using the evaluation device, it can be determined from the image capturing the final configuration E that there is a first region of interest 401, which is classified as a cable, and a second region of interest 402, which is a connecting pipe. By taking into account the relative position of the first and second regions of interest 401 and 402 for the classification, it can be determined whether the connecting pipe is correctly positioned in relation to the cable and whether the zero cut for the cable was carried out correctly.

Figure 6 shows a third exemplary final configuration E of a cable section. Figure 6 shows two cables K and two internal contact crimps IC arranged on the stripped cable ends AK, each also referred to as a contact pin. The final configuration E captured in the image is fed back to the evaluation device.

Four areas of interest 401 to 404 are determined by means of the evaluation device. The first and third areas of interest 401 and 403 correspond to the cable section in front of the stripped parts AK of the cable or the contact pins. The second and fourth areas of interest 402 and 403 correspond to an internal contact crimp IC. Each internal contact crimp IC is guided with one of its sections over a contact pin and crimped.

Using the evaluation device, it can be determined from the image capturing the final configuration E that there is a first and third area of interest 401, 403, which is classified as the cable section K adjacent to the contact pin AK, and a second and fourth area of interest 402 and 404, respectively, which is an internal contact crimp IC. These are again identified by corresponding frames R1 to R4. For reasons of clarity, the information in the frames has been omitted for Figure 6 waived.

By taking into account the relative position of the first and second regions of interest 401 and 402 or the relative position of the third and fourth regions of interest 403 and 404 for the classification, it can be determined whether the inner contact crimp was crimped onto the pin AK without errors. In particular, it can be determined from the distance of the first and second regions of interest 401 and 402 whether the position of the inner contact crimp to the pin is error-free. The same applies to the third and fourth regions of interest 403 and 404.

Figure 7 shows a fourth exemplary final configuration E. This comprises two cables K, each of which is equipped with a connector S. These cables K are to be equipped by means of a cable holding device H, here also referred to as a separator.

The image capture of this final configuration E is again carried out using an appropriately prepared evaluation device. In a first step, the region of interest 401 is determined again, which here corresponds to the holding device H. By means of the subsequent classification of the determined region of interest 401, it can be determined whether the holding device H is attached to the cable sections K without errors.

It is understood that the above examples of the application of the correspondingly designed neural network are not exhaustive, and a large number of other applications for cable processing are familiar to the person skilled in the art. In particular, an embodiment of the invention can also be used for checking the marking of Cables, in particular laser markings on the cable sheath, for the correct attachment of labels and tags to the cable, in particular with regard to compliance with the laser marking, the attachment of grommets and their holders, the attachment of couplings for sensor cables, the arrangement of protective caps on plugs, of ferrules on optical conductor end pieces, the arrangement of a silicone insulating tube on the cable, the attachment of outer conductors, etc.

Figure 7 shows a schematic process sequence as a flow chart for an embodiment of the method for operating a cable processing device.

In a method step S1, at least one cable section, preferably a plurality of cable sections, is processed from an initial configuration to a final configuration. This can be any processing step. With regard to the subsequent image capture of the at least one final configuration, the processing step should, however, influence the optical appearance of the cable end piece, since subsequent evaluation is carried out on an image basis.

In a method step S2, the at least one final configuration arranged in the recording area, preferably the plurality of final configurations, is recorded by means of an image recording device and the image is fed to an evaluation device for evaluating the final configuration recorded in the image. In the following, it is assumed - without limiting the applicability for processing only a single final configuration - that a plurality of final configurations are recorded. The image recording can be initiated by the evaluation device or a control device.

In a process step S3, the neural network is used to check whether all of the final configurations shown have been fully captured in the image or whether the image shows final configurations that have been captured incompletely. If final configurations are captured incompletely, an error signal is output, possibly only when the classification for the fully captured final configurations is available, from which incomplete image capture becomes apparent. The reason for the incomplete capture can then be eliminated.

For the fully recorded final configurations, a next step S4 uses the neural network to determine a plurality of regions of interest from the image, whereby each final configuration also contains a plurality of regions of interest. The neural network is trained to determine the desired number of areas of interest.

In a next method step S5, the determined areas of interest are classified with regard to the presence of a defect pattern and a result corresponding to the classification is determined.

In a next method step S6, a control signal is generated and output which is dependent on the at least one determined region of interest and/or dependent on the determined result of the classification of the at least one final configuration. The control signal is preferably output to a control device.

In a method step S 7, the information determined by the evaluation device from the captured image is graphically output on the basis of the control signal. The large number of determined regions of interest and an associated classification result are output on a monitor.

Furthermore, in a method step S8, the control device checks whether influencing the cable processing unit, a cable pre-processing unit or a cable subsequent processing unit, or separating a cable from production on the basis of the control signal is appropriate.

If necessary, a control intervention for a corresponding manipulated variable of the production takes place in a process step S9, which helps to avoid an error pattern for a final configuration, corrects the error pattern for a specific final configuration or causes a cable or cable section to be separated from production.

Since the devices and methods described in detail above are exemplary embodiments, they can be modified to a large extent by those skilled in the art in the usual way without departing from the scope of the invention. In particular, the mechanical arrangements and the size relationships of the individual elements to one another are merely exemplary.

LIST OF REFERENCE SYMBOLS

100

Cable processing device

200

Cable processing unit

201

Cable processing unit

202

Cable pre-processing unit

210

Cable holder

300

Image capture device

401

area of interest, first

402

area of interest, second

403

area of interest, third

404

area of interest, fourth

500

Evaluation device

600

Machine-readable program code

700

Control device

710

Cable production control

800

Image output device

K, K1, K2, K3, Kn

Cable sections

E, E1, E2, E3, En

Final configurations

C

Crimp sleeve

IC

Internal contact crimp

S

Plug

M

Coat

H

Cable holding device

R1, R2, R3, R4

Frames surrounding a determined region of interest

A1, A2, A3, A4

Classification information

S1

Simultaneous processing of a plurality of cable sections from an initial configuration to a final configuration by means of a cable processing unit.

S2

Capturing the multitude of final configurations arranged in the recording area

S3

Check for complete capture of final configurations

S4

Identification of a variety of areas of interest

S5

Performing the classification for the identified areas of interest

S6

Generating and outputting a control signal based on the identified areas of interest and their classification

S7

Graphical output of the results of the evaluation device

S8

Testing for control variable intervention for cable processing unit, cable pre-processing unit, cable subsequent processing unit and rejection

S9

Implementation of the measure determined from the examination

Claims (15)

Method for operating a cable processing device (100), comprising the method steps: - processing (S1) at least one cable section (K, K1, K2, K3, Kn)), in particular a cable end, by means of a cable processing unit (200) from an initial configuration to a final configuration (E, E1, E2, E3, En), - capturing (S2) at least one image of the at least one final configuration (E1, E2, E3, En) by means of an image recording device (300), - applying (S4, S5) a trained neural network to the at least one captured image, wherein the trained neural network is designed to determine (S4) at least one region of interest (401, 402, 403, 404) of the at least one final configuration (E, E1, E2, E3, En) from the at least one captured image in a first step and to carry out a classification of the at least one determined region of interest (401, 402, 403, 404) with regard to the presence of at least one learned error image of the at least one final configuration (E, E1, E2, E3, En) in a second step and to determine (S5) a result (A1, A2, A3, A4) associated with the classification, - generating and outputting (S6) a control signal depending on the at least one determined region of interest (401, 402, 403, 404) and/or depending on the determined result (A1, A2, A3, A4) of the classification of the at least one final configuration (E, E1, E2, E3, En). Method according to the preceding claim,
wherein an image is captured by means of the image capture device (300) which shows two to 100 final configurations, wherein the trained neural network is designed to determine (S4) at least one region of interest (401, 402, 403, 404) for the two to 100 final configurations (E, E1, E2, E3, En), in particular for each of the two to 100 final configurations (E, E1, E2, E3, En), and to determine (S5) a result (A1, A2, A3, A4) associated with the classification of the regions of interest (401, 402, 403, 404), and the trained neural network is applied to the captured image (S4, S5), wherein a control signal for the two to 100 captured final configurations (E, E1, E2, E3, En), in particular for each of the two to 100 captured final configurations (E, E1, E2, E3, En), based on the determined at least one region of interest (401, 402, 403, 404) and/or the result (A1, A2, A3, A4) associated with the classification for the respective final configuration (E, E1, E2, E3, En) and output (S6). Method according to one of the preceding claims,
wherein the trained neural network is designed to determine at least one first region of interest (401, 402, 403, 404) for at least one final configuration (E, E1, E2, E3, En) and at least one second region of interest (401, 402, 403, 404) that is spatially different from the first region of interest from the at least one captured image, and the neural network is further designed to additionally take into account a relative parameter, in particular a relative position, of the at least first region of interest (401, 402, 403, 404) and the second region of interest (401, 402, 403, 404) for the classification and to determine a result (A1, A2, A3, A4) associated with the classification, and wherein the trained neural network is applied to the at least one image (S4, S5), wherein a control signal for the at least one final configuration (E, E1, E2, E3, En) is generated and output (S6) based on the result (A1, A2, A3, A4) of the classification taking the relative position into account. Method according to one of the preceding claims,
wherein the capture (S3) of the image of the at least one final configuration (E, E1, E2, E3, En) is carried out by means of a digital image recording device (300), wherein the at least one region of interest (401, 402, 403, 404) is represented by at least 20 pixels, preferably more than 50 pixels. Method according to one of the preceding claims,
wherein the trained neural network is designed to determine (S3) the complete image capture of a final configuration (E, E1, E2, E3, En), wherein the trained neural network is applied to the captured image (S3), wherein in the case of the determined incomplete capture of at least one final configuration (E, E1, E2, E3, En) a control signal is generated and output which causes the notification of an image capture error. Method according to one of the preceding claims,
wherein by means of the control signal a graphic display of the at least one determined region of interest (401, 402, 403, 404) is generated in an image showing the at least one final configuration (E, E1, E2, E3, En), in particular the image to which the trained neural network was applied, in particular as a colored border (R1, R2, R3, R4) of the at least one determined region of interest (401, 402, 403, 404) on an image output device (800). Method according to one of the preceding claims,
wherein the control signal is used to cause a graphic display of the result (A1, A2, A3, A4) of the classification for the at least one final configuration (E, E1, E2, E3, En) in an image showing the at least one final configuration (E, E1, E2, E3, En), in particular the image on which the classification is based, on an image output device (800). Method according to claim 7,
whereby an error pattern assigned to the result (A1, A2, A3, A4) of the classification is reproduced in the form of an error pattern designation (A1, A2, A3, A4) by means of the graphic display. Method according to claim 8,
whereby a probability (A1, A2, A3, A4) for the determined error pattern is shown by means of the graphic display. Method according to one of the preceding claims,
wherein the operation of the cable processing unit (200) is influenced (S9) by means of the control signal in such a way that the final configuration (E, E1, E2, E3, En) of a cable section (K) to be processed subsequently is approximated to a desired final configuration (E, E1, E2, E3, En) by means of the cable processing unit (200). Method according to one of the preceding claims,
wherein by means of the control signal, a cable sequence processing of a cable sequence processing unit (201) downstream of the cable processing unit (200) of at least one classified final configuration (E, E1, E2, E3, En) is influenced depending on the result (A1, A2, A3, A4) of the classification of this final configuration (E, E1, E2, E3, En). Method according to one of the preceding claims,
wherein the control signal is used to cause a cable section (K) with a faulty final configuration (E, E1, E2, E3, En) to be rejected from a production process that is still to be completed, wherein rejection occurs when the result (A1, A2, A3, A4) of the classification for this final configuration does not match a predetermined Error pattern with a probability above a predetermined threshold value and cannot be corrected by means of a cable sequence processing unit (202). Cable processing device (100) comprising: - at least one cable processing unit (200) which is designed to receive at least one cable section (K), in particular a cable end, and to process it in such a way that the at least one cable section (K) is transferred from an initial configuration to a final configuration (E, E1, E2, E3, En), - an image recording device (300) arranged and designed to record at least one image of the at least one final configuration (E, E1, E2, E3, En), - an evaluation device (500) and a control device (700) for controlling and/or regulating the cable processing device (100), wherein the evaluation unit (500) is operatively connected to the image recording device (300) and the control device (700), and wherein machine-readable program code (600) can be loaded into the evaluation device (500) and/or into the control device (700), which, when executed, causes the method according to one of the preceding claims to be carried out. Machine-readable program code for an evaluation device and/or control device, which comprises control commands which, when executed by the evaluation device and/or the control unit, cause the method according to one of the preceding method claims to be carried out. Control and/or evaluation device with machine-readable program code which comprises control commands which, when executed by means of the evaluation device and/or control device, cause the method according to one of the preceding method claims to be carried out.

Images